[EM] Approval Voting vs Instant Runoff Voting

[Bart Ingles]

Buddha Buck wrote:
> 
> At 11:26 AM 02-15-2001 +1100, you wrote:
> >Forest,
> >
> >you wrote:
> >
> > >I would like to see that example.  You must have submitted it before my
> > >time.
> >
> >I posted it only a few days ago, but it was jammed in at the bottom of a
> >verbose message, so I'm not surprised if people didn't get to it.  Here it
> >is again (slightly reworded);
> >
> >This question is open to all the strategically minded posters; Say with the
> >above utilities (100, 25, 20, 0 - assign A,B,C,D respectively).  One opinion
> >poll shows (and, being an approval election, these are approval polls so
> >they show the predicted winner), that A will get 36 percent of the vote, B
> >will get 40, C will get 45 and D will get 44.
> >
> >You even have information about the accuracy of the polls; accuracy within
> >5% of the predicted outcome is 90% (that is, 90% chance that A will get
> >between 31 and 41%), and, our general understanding of such polls claiming
> >this kind of accuracy is that it is more likely to be closer to the
> >predicted outcome than further away.  How should you vote?
> 
> I'm not up to date on all the strategies involved, but...
> 
> Without any strategy, based on my utilities, I'd probably vote approval on
> "A" and no approval on the others.  A looks so much better than the rest to
> me that I see nothing to be gained by the others.  B, the closest
> competator, is so far down from A that I'd not really be satisfied if B won
> instead of A.
> 
> However, with the poll results, I've a quandry.  A looks like he has a very
> low chance of winning, and the real race is between C and D.  Even though C
> isn't very good, he is (to me) much, much, MUCH better than D -- who I
> don't like at all.  As such, I feel I'd have to throw my support (and vote)
> towards C so that he has a better chance of defeating D.  Since this is
> approval, there is no reason for me to not also vote for my more preferred
> options.  So I'd vote for A, B, and C and against D.

Good point -- I hadn't noticed that A was placed that much lower in the
polls than either C or D.  So it probably is a bit of a toss-up to
decide which of the admissible strategies to use.  And the admissible
strategies in this case are [A], [AB], [AC], and [ABC].

But so what?  All voting systems have similar toss-up situations.  With
ranked systems, these situations can be considerably worse than
approval's.


One thing we're ignoring is the usual spread of voter opinions.  In some
cases, voter opinions of B and C will be high enough that they will
gladly vote for two or three candidates; in others, they will bullet
vote regardless of the polling data.  These two groups will tend to
bracket the truly ambiguous voters, thereby constraining the number of
compromise votes between some minimum and maximum levels.  This would
tend to dampen any effect of ambiguous voters somehow stampeding in one
direction or another.

With ranked systems, this voter opinion spread can have the opposite
effect -- actually making things less predictable.  This would be a
problem if fringe groups engage in order-reversal tactics.

Bart